CN104093354A - Method and apparatus for assessment of medical images - Google Patents

Abstract

A method of determining the degree of uptake of a PET maker in an individual candidate PET scan is provided. The method includes the steps of: (a) calculating a series of representative matched controlled PET and MRI templates for a series of controlled sample scans of individuals; (b) computing a series of brain surfaces from the matched templates; (c) aligning the individual candidate PET scan with the candidate templates; (d) aligning the candidate PET images with the series of brain surfaces; (e) selecting a predetermined (M) best candidate templates for each surface location based on a similarity measure between the candidate PET values and the corresponding controlled PET scans; (f) computing M weights for each surface location, utilizing a corresponding MRI tissue map; and (g) utilizing the M weights to combine a corresponding M template tissue indicators from corresponding MRI templates into an average brain surface indicator.

Description

For assessment of the method and apparatus of medical image

Technical field

The present invention relates to understand and process the field of medical image, and disclose especially for understanding at the labelling occurring such as the image of pet art (PET) image or single photon emission computerized tomography,SPECT image (SPECT), for detecting abnormal phenomena.

Background technology

In description, to any discussion of background technology, should not to be considered to this type of technology be widely understood or the part that forms general knowledge known in this field.

Utilizing is an important process such as the in-vitro diagnosis disease that is imaged on of PET or SPECT imaging.Particularly for the situation such as the such neurodegenerative disease of senile dementia (AD).Important to the deciphering of suitable image for the understanding and the discriminating aspect that improve disease treatment.

Prior art comprises many for obtaining and analyze the system of PET and other imaging patterns.For example, Liljat etc. in the open WO2008/093057 of PCT, disclose a kind of such in investigation AD for analyzing the system of PET and other images.Chen Kewei etc. discloses another system in U.S. Patent Application Publication No. .2006/074290.In all this systems, it is desirable to provide the diagnostic message about image state rapidly.

Amyloid-beta (A β) speckle is the most general in the pathological characters of senile dementia (AD), and this speckle may be in dementia by just having occurred before being diagnosed perhaps many years ago.The immediate development of functional image agent makes it can be evaluated at the amyloid beta deposition in live body.A kind of promising known radioactive indicator be B-type Pittsburgh complex ( ¹¹c-PiB), this material can combine with high-affinity and high specific and A amyloid beta.Demonstrate AD patient and trended towards thering is higher PiB in cortex region than normal control.Other important amyloid imaging compounds are developed (such as Florbetapir), and also can use in the present invention.

The picked-up by histological types (grey matter, white matter and CSF) can be so that diagnosis and monitoring be dull-witted in cortex region for assessment PiB.But due to the low resolution of PET image and lack structural information, present method conventionally relies on parallel MRI image and determines tissue regions.Further, the binding pattern of labelling often has nothing to do with potential organizational structure and may be alterable height.Be included on MRI image tissue segmentation (segmentation) simultaneously and for the multimode registration (multi-modular registration) between MRI and the PET image of each main body such as those disclosed general process in Lilja etc.

Although this type of estimation is relatively accurate, due to clinical setting and because a variety of causes (as, claustrophobia, metal are implanted, etc.) and lack MRI scanning, the appraisal procedure that is therefore independent of MRI is desirable.

Summary of the invention

The object of this invention is to provide an improved form of the assessment of the labelling picked-up in the medical image such as PET or SPECT image.

According to a first aspect of the invention, provide a kind of for determine the method for the picked-up degree of PET labelling in indivedual candidate PET scanning, the method comprises the following steps: (a) calculate the PET of a series of representative match controls and the MRI template scan sample for a series of controls of individuality; (b) a series of brain surface of formwork calculation from mating; (c) adopt candidate template to aim at indivedual candidate PET scanning; (d) a series of brain surface described in employing candidate PET image alignment; (e) similarity measurement between the scanning of the PET based on candidate PET value and corresponding control selects predetermined (M) optimal candidate template for each surface location; (f) utilize corresponding MRI to organize the similarity measurement between mapping graph and candidate PET and PET template, calculate M weight for each surface location; (g) utilize a described M weight to organize index to combine the picked-up of the candidate PET of each position of estimating average brain surface's index a corresponding M template.

The method is preferably further comprising the steps of: (h) the average brain surface of combination and candidate PET scan-data create the average brain surface of the combination for showing.In some embodiments, step (c) can be included in to adopt in described a series of candidate template aligning candidate PET image and utilize candidate CT or X-ray scanning data.

According to a further aspect in the invention, provide a kind of for be specified to the method for the picked-up degree of image scale note in indivedual candidate's imaging mark scans, the method comprises the following steps: (a) calculate the imaging mark scan of a series of representative match controls and the tissue mark's template scan sample for a series of controls of individuality; (b) according to delimiting surface in a series of body of formwork calculation of coupling; (c) adopt candidate template to aim at indivedual candidate's imaging mark scans; (d) adopt delimitation surface in described a series of body to aim at described indivedual candidate's imaging mark scans; (e) similarity measurement based between candidate's imaging mark value and the corresponding imaging mark scan of controlling to select predetermined (M) optimal candidate template for each surface location; (f) utilize corresponding tissue mark mapping graph and similarity measurement, calculate M weight for each surface location; (g) utilize a described M weight to organize index to be bonded in average brain surface's index from corresponding M template of corresponding tissue templates.

Imaging mark scan can comprise pet art (PET) scanning or single photon emission computerized tomography,SPECT image (SPECT) scanning.Each tissue mark template can be calculated according to having the image of different subjects of known image attributes.Multiple tissue templates can be according to have the candidate of known image attributes and the similar features of main body from selecting from wider template set.Template and candidate image are preferably divided.

According to a further aspect in the invention, a kind of method of determining large brain capture imaging labelling in subject image is provided, the method comprises the following steps: prepare a series of brain templates according to the sample image of controlling, this template comprises the sample image of a series of controls of common registration, and create probability mapping graph (mapping), the estimation that this mapping graph comprises the probability that specific collection of illustrative plates (atlas) voxel contains grey matter; From affiliated brain template, determine the respective surfaces between grey matter interface and white matter interface; For candidate's subject image, described candidate image is mapped to corresponding brain template; And for the gray area part of described brain template, the corresponding relation that shines upon described subject image and carry out grey matter picked-up.The sample image of controlling can comprise PET and MR image the two.

Corresponding relation is to measure with respect to surperficial predetermined direction or volume between grey matter and white matter.Mapping can betide multiple templates.Utilize Bayesian network or weighted sum or voting rule or other integration technologies to come preferably in conjunction with multiple mappings.

Brief description of the drawings

The preferred embodiment of the present invention is only described in the mode of example with reference to the accompanying drawings, wherein:

Fig. 1 schematically shows a kind of form of the operating environment of carrying out preferred implementation;

Fig. 2 shows the Part I of the flow chart of the step of preferred implementation;

Fig. 3 shows the Part II of the flow chart of the step of preferred implementation;

Fig. 4 shows the process of template registration;

Main body is mapped to the mapping result of template or the diagram of collection of illustrative plates value by Fig. 5 to Fig. 8; Wherein Fig. 5 has shown the multichannel chromatogram method of method based on average correlation coefficient and the contrast of free hand drawing spectral method that utilize dependence MRI.This multichannel chromatogram method demonstrates for 104 nearly all test subjects and produces all the time higher dependency.

Fig. 6 shows the multichannel chromatogram method of mean error and the contrasting of free hand drawing spectral method on the each summit based between PET and the method for dependence MRI.All the time produce lower error for nearly all this multichannel chromatogram method of 104 test subjects.

Fig. 7 shows the average ROI fixed point multichannel chromatogram method of error and the contrasting of free hand drawing spectral method based between PET and the method for dependence MRI.All the time produce low error for this multichannel chromatogram method of nearly all ROI.

Fig. 9 shows the example that AD patient is shone upon; And

Figure 10 shows the example that normal patient is shone upon.

Detailed description of the invention

In embodiments of the present invention, provide the method for " only PET image ", the method is averaging to the PET image of main body and within the scope of ROI area-of-interest (ROI) atlas registration to PiB picked-up value.Wherein, in the situation that only comprising free hand drawing spectrum, the estimated accuracy of the method may depend on selection and the registration error of collection of illustrative plates.Have been found that by utilizing multichannel chromatogram can obtain improved result.Embodiment provides more " only PET " method of high robust, and the method has been improved the posterior probability of assessment.In order to reach this object, adopt following three kinds of main strategies:

1. organize probability mapping graph (map) to be introduced into instruct the measurement of PiB value, replace the hard ROI subregion copying from the ROI collection of illustrative plates of single distortion.Probability mapping graph is according to the priori of training set study, therefore combines Population Variation and has stronger robustness than the ROI obtaining from single collection of illustrative plates image simply.

2. many collection of illustrative plates are used to offset the registration error from single collection of illustrative plates.The utilization of multichannel chromatogram also allows each PET image from different PET collection of illustrative plates.The use of multiple PET collection of illustrative plates allows the selection of each position of the main body in data base similar.In other words, the method for second illustrative embodiments can be used in given patient, for template A, B, C and template C, D, the E for a pixel on temporal lobe of a pixel in brain frontal cortex.

The optimal subset of the specific collection of illustrative plates of main body can be selected from a large amount of collection of illustrative plates, and carrys out the posterior probability of improved estimator in conjunction with Bayes (Bayesian) framework.Weight in conjunction with different templates can change individual element.PiB picked-up value then directly estimated, this has been avoided the significant need that grey matter is cut apart.Therefore, use multiple collection of illustrative plates to reduce statistically and appear at registration and the sampling error in each independent collection of illustrative plates.

3. the probability mapping graph based on colony except obtaining from collection of illustrative plates, the specific probability mapping graph of main body also can be cut apart and is utilized by the non-local mean of the main consuming body PET image.This also provides the method for the weight for calculating each pixel.Utilize multiple collection of illustrative plates, derived the probability mapping graph that is used in particular for special body this cutting apart based on community information, and thereby further improved priori and posterior estimation.

Compare with art methods (prior art needs MRI and PET image to come for accurately estimating the PiB picked-up in grey matter conventionally), embodiments of the present invention can only be utilized PET image, but still also provide quite accurately and to estimate, this estimation only has very little estimation difference and has with the method based on MRI and has high dependency.Although the method proposing for having general application based on volume and the assessment of the PiB based on surperficial, first the measurement based on surperficial is described.Measurement based on volume is relatively simple and can in the situation that not using surface model, carry out.

The method that embodiments of the present invention provide can be used as clinical examination instrument for diagnosis and monitoring senile dementia.But the method can be applied to the labelling of other PET of amyloid (AV1), and there is enough generality and be applied to any other PET labelling of any pathology and other SPECT labellings of any pathology.

First forward Fig. 1 to, the figure illustrates the operating environment 1 for implementing preferred implementation.In this environment, PET image is scanned 2 and processes to be stored in data base 3.Image analysis system 4 is carried out the image analysis method of preferred implementation and is exported the image report 5 that comprises multiple measurements.

Preferred implementation can be proceeded following step, as at first as shown in Fig. 2 and Fig. 3:

At first, a series of N (wherein N equals 20 in one embodiment) expression property template is formed, for having the test subject of known AD degree.For each test subject, PET and corresponding MRI image are acquired and aim at.And in the situation that CT scan can be used, these images can also be by coupling for test subject.According to each PET and optional CT scan, corresponding brain surface is calculated.

Template is stored in data base 21.

New main body then presents with the form 24 of PET scanning and optional CT scan (as available).

Preferred implementation is then carried out following steps:

Step 1: scan and utilize optional CT scan (available in the situation that) based on PET, new main body PET aims at N template.

Locate with respect to main body PET step 2:N template brain surface's position.Each in N template surface is regarded as surface mesh.For the each point on surface mesh, new main body PET value 26 with between PET value 27,28 corresponding to each template place, contrast, and obtain difference measure.

Step 3: for each surface location, M optimal Template mated the similarity between main body PET based on new and corresponding template PET and be stored.

Step 4:M optimal Template coupling is used to leading-out needle to the M of an each surface location weight.Utilize template to organize mapping graph, M weight be used to merge or the corresponding PET value of weighting to produce overall PET value.Merge taking position one by one as basis and occur.

Step 5: it is upper that result is displayed on " average brain surface ", wherein average brain surface derives by weighting template surface.And PET value is displayed on obtained average brain surface 36.Because all brain template surfaces are by common registration, therefore candidate PET can be presented in any one of template surface or on average surface to represent population.

the utilization of optional CT imaging

Optional at another of above embodiment 'sin refinement, should be noted that and in the PET/CT scanner that current medical practice usually promotes in combination, produce together the scanning of CT type and PET scanning.Scan available in the situation that at these, they can be used to registration process ideally.Although CT scan does not provide good tissue to delimit, it provides the delimitation of extraordinary liquid and skeleton.Can be used for new main body in CT scan, CT scan can be utilized with following various ways:

1. initial, in step 1, CT scan can be bonded in the aiming at of corresponding PET scanning and template.CT scan can provide the skeleton of clearly new main body to delimit and therefore can be used to alignment procedures.

In addition, in step 4, suppose that patient's CT can use, because CT delimit brain surface's border condition, therefore can be for estimating the position of grey matter boundary.Therefore CT can be used to revise the grey matter calculating of current PE T scanning in borderline region, and thereby improves totally and estimate.

Although having described before the operation of preferred implementation, in the time implementing described below invention, multiple processing refinements and change are available for a person skilled in the art.

1. initial's figure pretreatment template establishment: first all PET imagings of catching can carry out pretreatment.This comprises the brightness normalization of all PiB PET images, and the preparation of brain template.MRI scanning is assumed to be and only can be used for template.

For template data set, be necessary the surface of the boundary between pretreatment MRI image, PET image and grey matter and white matter, or the surface between grey matter and CSF alternatively.

First, each MRI image is spatially standardized as collection of illustrative plates (collection of illustrative plates using in the exemplary embodiment, is Collins collection of illustrative plates: Collins, D., Zijdenbos, A., Kollokian, V., Sled, J., Kabani, N., Holmes, C., Evans, A, 1998.Design and construction of a realistic digital brain phantom.IEEE Trans.Med.Imag.17 (3), 463-468).The prior probability mapping graph of the Main Tissues (GM, WM and CSF) being associated with collection of illustrative plates is also utilized (provides part to have the SPM of collection of illustrative plates in our example, Ashburner, J., Friston, K., 1999 " Nonlinear spatial normalization using basis functions " Hum.Brain Mapp.7 (4), 254-26).Collection of illustrative plates is alternatively changed places and is calculated from arbitrary image data storage capacity with the priori being associated.

MRI image and PET image are by local rigid transformation and by common registration.The method using can be to be disclosed in Ourselin, S., Roche, A, Subsol, G., Pennec, X., Ayache, N., 2001.Reconstructing a3D structure from serial histological sections.Image Vis.Comput.19 (1), 25-31.

After registration, the brightness value of PET image carries out standardization (Lopresti BJ by the picked-up value ratio (SUVR) of standard, Klunk WE, Mathis CA, Hoge JA, Ziolko SK.Lu X etc., " Simplified quantification of pittsburgh compound B amyloid imaging PET studies:a comparative analysis ", J Nucl Med2005; 46:1959-72Lopresti etc., 2005) be used to ensure comparison between main body and the comparison of body interior.The region that SUVR is defined as comprising specific binding is with respect to the value in the region that does not comprise specific binding.Because little ectocinerea is considered to there is no senile plaque, therefore little ectocinerea is usually used as the region (Joachim that does not comprise specific binding of reference, C., Morris, J., Selkoe, D., 1989.Diffuse senile plaques occur commonly in thecerebellum in Alzheimer's disease.American J.Pathol.l35 (2), 309-319).Be used to make the region in PET image to localize for standardization from the cerebellum mask (cerebellum mask) of MRI.Finally, the interface surface between grey matter and white matter is extracted for each collection of illustrative plates from the MRI image of cutting apart.

This pre-treatment step can be as following in being set forth in: Fripp J., Bourgeat P., Acosta0., Raniga P., Modat M., Pike E., Jones G., O'Keefe G., Masters L., Ames D., Ellis A., Maruff P., Currie J., Villemagne L., Rowe C., Salvado O., Ourselin S., 2008.Appearance modeling of11C PiB PET images:characterizing amyloid deposition in Alzheimer's disease, mild cognitive impairment and healthy aging.Neuroimage43 (3), 430-439.

Template preparation plays initial set with base image and proceeds, and this plays initial set can comprise a series of corresponding MRI and the PET scanogram for a collection of main body.

A series of independent templates can be created.For each template, their MRI and PET image are by common registration (aligning) rigidly.Be organized on template MRI image dividedly, and then, grey matter/white matter interface is determined.In order to make user can check inner cortex region, the surface at grey matter/white matter interface is separated into left hemisphere and right hemisphere.And each template establishment has grey matter probability mapping graph, this grey matter probability mapping graph has indicated image voxel to belong to the probability of grey matter.

Grey matter probability mapping graph can create by using at gauss hybrid models (the Gaussian Mixture Model) dividing method of standard based on brightness of the description of the Fripp quoting before etc.

2. template surface registration: multiresolution EM-ICP method is applied in the different templates surface at grey matter/white matter interface and sets up corresponding relation.After surface registration, template surface by resampling to have the respective vertices of equal number.EM-ICP method can be as following in being set forth in: Granger, S., & Pennec, X. (2002), " Multi-scale EM-ICP:A Fast and Robust Approach for Surface Registration ", Computer Vision-ECCV2002, 2353, 418-432Springer, and Combes, B.and Prima, S., 2010 " An efficient EM-ICP algorithm forsymmetric consistent non-linear registration of point sets " Medical Image Computing and Computer-Assisted Intervention (MICCAI), 594-601.

3. the registration to PET image: affine/slight deformable registration between the PET of special body image and corresponding template is performed to take surface and probability mapping graph to principal space from template space.

The process of affine registration illustrates by Fig. 3.

Alternatively, come the probability mapping graph of conversion of self-template in the case of having substituted use, the probability mapping graph in principal space can be cut apart generation by the non-local mean on main body PET image.

4. merge by multi-template the measurement based on surperficial of carrying out: average grey matter PIB picked-up along the reference direction on the collection of illustrative plates surface transforming by independently by transform accordingly organize the guiding of probability mapping graph and measured.Then,, according to the local weighted scheme of the local similarity between collection of illustrative plates PET image and main body PET image based on transforming, measure in conjunction with Bayesian network.Similarity can be measured by standardized common information or other indexs.

other weighting schemes can be used.For example, M weight can be used to weighted average in the situation that organizing index in conjunction with corresponding template.Another alternative technology is organized index for what provided by voting algorithm for definite corresponding template.

As further substituting, in order to merge the result obtaining from selected multiple templates, Bayesian frame can as described belowly be utilized.

Given PET image I (x), wherein x presentation video voxel, target is along the collection of illustrative plates surface S transforming ^treference direction measure the average PiB picked-up in grey matter.This equals to estimate expected value E _{x ∈ Δ}[δ (I, x, l)], wherein δ (I, x, l) is target function, it is defined as follows:

Symbol Δ represents the intersection of the line of the reference direction of summit ν and PET image I surfacewise.Symbol l, for organizing label, represents GM, WM and CSF by value 1,2,3 respectively.Expected value can be estimated as follows:

$\begin{array}{c}{E}_{x\∈\mathrm{\Δ}}\left[\mathrm{\δ}\right(I,x,l\left)\right]=\underset{x\∈\mathrm{\Δ}}{\∫}\mathrm{\δ}(I.x,l)p(I,x,l)\mathrm{dx}\\ =\underset{x\∈\mathrm{\Δ}}{\∫}\mathrm{\δ}(I,x,l)p\left(l\right|I,x\left)p\right(I,x)\mathrm{dx}\\ ={\∫}_{x\∈\mathrm{\Δ}}I\left(x\right)p(l=1|I,x\left)p\right(I,x)\mathrm{dx}\end{array}.---\left(1\right)$

Consider discrete probability, obtain following result:

$E_{x\∈\mathrm{\Δ}}\left[\mathrm{\δ}\right(I,x,l\left)\right]={\mathrm{\Σ}}_{x\∈\mathrm{\Δ}}I\left(x\right)P(l=1|I,x)P(I,x).---\left(2\right)$

Suppose that x is equably from Δ sampling, probability wherein | Δ | be the length of Δ.By marginalisation joint probability the template of posteriority marking probability P (l|I, x) from transforming (i=1 ... n, the quantity that wherein n is selected template) estimate:

$P(\backslash |I,x)={\mathrm{\Σ}}_{i=1}^{n}P(l,A_i^T|I,x)={\mathrm{\Σ}}_{i=1}^{n}P\left(l\right|A_i^T,I,x)P\left(A_i^T\right|I,x).---\left(3\right)$

Here be illustrated in the template of conversion in, voxel x is as the probability of GM, this template can from the template probability mapping graph transforming, obtain.Probability measure the template that voxel x can be aligned in image I and conversion well between probability.In our method, be set to the inverse of the measurement of the standardized common information of partial estimation in the contiguous N of x (x) proportional.Namely, due to the low resolution of PET image, the size of N (x) should be too not little, otherwise the common information obtaining is by matching noise.In the measurement of having carried out, N (x) is set to 30 × 30 × 30, and this has covered all voxels along line Δ.Therefore, with respect to variable x (x ∈ Δ), for constant.

In conjunction with (2) and (3), obtain:

$\begin{array}{c}{E}_{x\∈\mathrm{\Δ}}\left[\mathrm{\δ}(I,x,l)\right]=\frac{1}{\left|\mathrm{\Δ}\right|}{\mathrm{\Σ}}_{x\∈\mathrm{\Δ}}I\left(x\right)P(l=1|I,x)\\ =\frac{1}{\left|\mathrm{\Δ}\right|}{\mathrm{\Σ}}_{x\∈\mathrm{\Δ}}\left(I\left(x\right)\underset{i=1}{\overset{n}{\mathrm{\Σ}}}P\right(l=1|A_i^T,I,x)P\left(A_i^T\right|I,x))\\ =\underset{i=1}{\overset{n}{\mathrm{\Σ}}}P\left(A_i^T\right|I,N\left(x\right))\left(\frac{1}{\left|\mathrm{\Δ}\right|}{\mathrm{\Σ}}_{x\∈\mathrm{\Δ}}I\left(x\right)P(l=1|A_i^T,I,x)\right)\end{array}---\left(4\right)$

Equation (4) has shown the adeditive attribute retaining for estimation average PiB at each surface vertices place: can be by realizing according to the independent estimations of each single collection of illustrative plates and then carrying out linear combination in the mode of weighting according to the estimation of multiple collection of illustrative plates.In conjunction with weight reflect the template of test pattern I and conversion between aligning.Assessed by Local Metric owing to aiming at, therefore this combination is nonlinear for whole surface.This adeditive attribute contributes to the method being adopted in the time that template set need to dynamically be determined.By variation is only limited to affected template, this makes easily to change selected template.

According to single template while carrying out individuality estimation, by it, the probability as grey matter voxel is weighted PiB value I (x), namely, this implicit difiinition there is the grey matter region on soft border, viewed variation in training population has been reflected on this soft border.Therefore, cut apart with respect to the disclosed hard grey matter of prior art, the estimation proposing can improve the robustness of registration error.

Use multiple templates to be found to improve the estimation of carrying out according to individual template.The improved example obtaining in test process is as shown in Fig. 5 to Fig. 8.

5. surface is visual: once the measurement based on surperficial is performed, and so, for each main body, left hemisphere and right hemisphere are combined and manifested from eight angles.Can automatically grab screen, this is conducive to vision-based detection or produces further report.

The example results of an AD and a normal control is respectively shown in Fig. 9 and Figure 10.

explain

The following description and drawings utilize reference number to assist the 26S Proteasome Structure and Function of understanding embodiment.Identical reference number is used to different embodiments and specifies the feature with same or similar function and/or structure.

Accompanying drawing need to be regarded as an entirety and to text combination relevant in this explanation.Especially, some accompanying drawings have optionally been ignored all features in all examples special characteristic more clearly to provide a description.Although this contributes to reader, it is unexposed or do not need for the operation of relevant embodiment that it should not be considered to these features.

In whole description, " embodiment " or " embodiment " mean special characteristic, structure or the characteristic relevant with being included in embodiment at least one embodiment of the present invention.Thereby, occur that in each position of whole description phrase " in one embodiment " or " in embodiment " not necessarily relate to identical embodiment, but this is also also passable certainly.In addition, special characteristic, structure or characteristic can combinations in any suitable manner in one or more embodiments, as it will be apparent to those skilled in the art according to the disclosure.

Similarly, should be understood that in the description of above illustrative embodiments of the present invention, each feature of the present invention is combined in single embodiment, accompanying drawing or its explanation simultaneously, makes the object of disclosure smoothness and reaches auxiliary one or more the object of understanding in each inventive aspect to reach.But disclosed method can not be interpreted as reflecting that desired invention need to be than knowing the purpose with more feature of describing in each claim.On the contrary, as the following claims reflect, the aspect of innovation is fewer than all features of above independent disclosed embodiment.Thereby along with the claim illustrating illustrates and combines with this accordingly, and each claim self is as independent embodiment of the present invention.

In addition, although embodiments more described here comprise that (but non-other) comprises feature in other embodiments, but the combination of the feature of different embodiments should be understood within the scope of the present invention, and those skilled in the art are to be understood that this has formed different embodiments.For example, in following claim, any embodiment in desired embodiment can be used in any combination.

In addition, the combination that some embodiments are described to method or method key element in this, this can realize by the processor of computer system or by other devices of carrying out function.Thereby the processor with the necessary instruction of the key element of carrying out these class methods or method has formed the means that require for carrying out the method or method.In addition, the key element in this description of equipment embodiment is the example of the device for carrying out the function of being carried out by key element, carries out object of the present invention to reach.

In the explanation providing at this, many specific detail are described.But, should be understood that embodiments of the present invention can be implemented in the situation that there is no these specific detail.In other examples, for fear of obscuring understanding of the present invention, known method, structure and technology are not specifically shown.

Similarly, should know, the term of the coupling that used in the claims should not be interpreted as only limiting to direct connection.Term " coupling " can be used with " connection " and derivative term.Should be understood that these terms are not synonym each other.Thereby, express scope that device A is coupled to equipment B and should not be limited to the wherein output of device A and be connected directly to equipment or the system of the input of equipment B.It means the path existing between the output of A and the input of B, and this path can be the path that comprises other equipment or device." coupling " can mean the contact of two or more element direct physical or electrically contact, or two or more element directly contact but still co-operate or influence each other mutually each other.

Thereby, although having described, which is considered to the preferred embodiment of the present invention, those skilled in the art are to be understood that, in the situation that not deviating from spirit of the present invention, can it be carried out other and further be revised, and should be appreciated that change and revise all and fall within the scope of the present invention.For example, the formula more than providing has only represented operable process.Can from block diagram, increase or delete function, and can be between functional module swap operation.Can increase or delete step from described method in the scope of the invention.

Claims (17)

1. for determine a method for the picked-up degree of PET labelling in indivedual candidate PET scanning, the method comprises the following steps:

(a) calculate the PET of a series of representative match controls and the MRI template scan sample for a series of controls of individuality;

(b) a series of brain surface of formwork calculation from mating;

(c) adopt candidate template to aim at indivedual candidate PET scanning;

(d) a series of brain surface described in employing candidate PET image alignment;

(e) similarity measurement between the scanning of the PET based on candidate PET value and corresponding control selects predetermined (M) optimal candidate template for each surface location;

(f) utilize corresponding MRI to organize mapping graph, calculate M weight for each surface location;

(g) utilize a described M weight to organize index to be bonded in average brain surface's index from corresponding M template of corresponding MRI template.

2. method according to claim 1, further comprising the steps:

(h) create the average brain surface for showing in conjunction with template brain surface.

3. according to the method described in above claim arbitrarily, wherein step (c) is further included in to adopt in described a series of candidate template aligning candidate PET image and utilizes candidate CT or X-ray scanning data.

4. for be specified to a method for the picked-up degree of image scale note in indivedual candidate's imaging mark scans, the method comprises the following steps:

(a) calculate the imaging mark scan of a series of representative match controls and the tissue mark's template scan sample for a series of controls of individuality;

(b) according to delimiting surface in a series of body of formwork calculation of coupling;

(c) adopt candidate template to aim at indivedual candidate's imaging mark scans;

(d) adopt delimitation surface in described a series of body to aim at described indivedual candidate's imaging mark scans;

(e) similarity measurement based between candidate's imaging mark value and the corresponding imaging mark scan of controlling to select predetermined (M) optimal candidate template for each surface location;

(f) utilize corresponding tissue mark mapping graph, calculate M weight for each surface location;

(g) utilize a described M weight to organize index to be bonded in average brain surface's index from corresponding M template of corresponding tissue templates.

5. method according to claim 4, wherein said imaging mark scan comprises pet art PET scanning or single photon emission computerized tomography,SPECT image SPECT scanning.

6. method according to claim 4, wherein each tissue mark template is calculated according to having the image of different subjects of known image attributes.

7. method according to claim 6, wherein said multiple tissue templates are that basis has the candidate of known image attributes and the similar features of main body is selected from wider template set.

8. according to the method described in above claim arbitrarily, wherein said template and candidate image are divided.

9. for determine a method for large brain capture imaging labelling in subject image, the method comprises the following steps:

Prepare a series of brain templates according to the sample image of controlling, this template comprises the sample image of a series of controls of common registration, and creates probability mapping graph, the estimation that this mapping graph comprises the probability that specific collection of illustrative plates voxel contains grey matter;

From affiliated brain template, determine the respective surfaces between grey matter interface and white matter interface;

For candidate's subject image, described candidate image is mapped to corresponding brain template; And

For the gray area part of described brain template, the corresponding relation that shines upon described subject image and carry out grey matter picked-up.

10. method according to claim 9, the sample image of wherein said control comprise PET and MR image the two.

11. methods according to claim 9, wherein said subject image is PET image.

12. methods according to claim 9, wherein said corresponding relation is that predetermined direction or the volume of the surface location between with respect to grey matter and white matter or between grey matter and CSF measured.

13. according to the method described in any claim in claim 9 to 12, and wherein said mapping betides multiple templates.

14. methods according to claim 13, wherein said multiple mappings utilize Bayesian network combination.

15. according to the method described in claim 1 or 4, and wherein said step (g) comprises organizes index as weighted average for the corresponding template of combination by a described M weight.

16. according to the method described in claim 1 or 4, and wherein said step (g) comprises organizes index for voting algorithm to determine corresponding template by a described M weight.

15. 1 kinds for implementing the claims the equipment of the method described in any claim of 1 to 14.